- Email: [email protected]
Characterization of recombinant xylitol dehydrogenase from Galactocandida mastotermitis expressed in Escherichia coli
Chemico-Biological Interactions 143 /144 (2003) 533 /542 www.elsevier.com/locate/chembioint
Characterization of recombinant xylitol dehydrogenase from Galactocandida mastotermitis expressed in Escherichia coli Bernd Nidetzky a,*, Heidemarie Helmer a,b, Mario Klimacek a, Regina Lunzer a,b, Gerhard Mayer a,b b
a Institute of Biotechnology, Graz University of Technology, Petersgasse 12/I, A-8010 Graz, Austria Institute of Food Technology, University of Agricultural Sciences Vienna, Muthgasse 18, A-1190 Vienna, Austria
Abstract The plasmid-encoded gene of xylitol dehydrogenase from the yeast Galactocandida mastotermitis was expressed in Escherichia coli at 25 8C. Recombinant enzyme was isolated in 70% yield using two steps of biomimetic affinity chromatography with the dye ligand Procion Red HE3B immobilized onto Sepharose 4B-CL. Similar to natural enzyme, recombinant xylitol dehydrogenase is a functional homotetramer with a stoichiometric content of catalytic zinc in each 37-kDa subunit. Though lacking bound Mg2 found in xylitol dehydrogenase isolated from yeast cell extracts, the recombinant enzyme is as active and stable as the native enzyme. Stereospecificity of enzymic hydrogen transfer from NADH has been determined by 1H-NMR and is 4-pro-R . A detailed steady-state kinetic analysis has been carried out for the enzymic reaction, polyol/NAD l/ketose/NADH/H , at pH 7.5 and 25 8C using xylitol and Dxylulose, the physiological polyol /ketose pair, as well as D-sorbitol and D-fructose. Primary deuterium kinetic isotope effects on steady-state kinetic parameters for oxidation of D-sorbitol and reduction of D-fructose have been measured at pH 7.5. Combined results of initial-rate analysis and isotope effect studies suggest that the enzyme utilizes a preferentially ordered kinetic mechanism in which NAD binds before D-sorbitol and D-fructose is released before NADH. Dissociation of NADH appears to be the main rate-limiting step for D-sorbitol oxidation under substratesaturated reaction conditions. # 2002 Elsevier Science Ireland Ltd. All rights reserved. Keywords: Xylose metabolism; Medium-chain dehydrogenase/reductase; Kinetic mechanism; Stereospecificity
1. Introduction
Abbreviations: GmXDH, recombinant XDH from Galactocandida mastotermitis expressed in Escherichia coli ; KIE, primary deuterium kinetic isotope effect; MDR, medium-chain dehydrogenase/reductase; XDH, xylitol dehydrogenase; XR, xylose reductase. * Corresponding author. Tel.: /43-316-873-8400; fax: /43316-873-8434. E-mail address: [email protected] (B. Nidetzky).
Xylose reductase (alditol:NAD(P) -1-oxidoreductase; XR; EC 1.1.1.21) and xylitol dehydrogenase (xylitol:NAD -2-oxidoreductase; XDH; EC 1.1.1.9) constitute the initial catabolic pathway for D-xylose in yeasts and fungi. The net reaction of this pathway is an isomerization of D-xylose into D-xylulose which is then phosphorylated at the 5-OH and metabolized further via pentose
0009-2797/02/$ - see front matter # 2002 Elsevier Science Ireland Ltd. All rights reserved. PII: S 0 0 0 9 - 2 7 9 7 ( 0 2 ) 0 0 2 1 5 - 6
534
B. Nidetzky et al. / Chemico-Biological Interactions 143 /144 (2003) 533 /542
phosphate and Embden /Meyerhof /Parnas pathways [1]. XDH is responsible for the oxidation of xylitol, resulting from the XR-catalyzed reduction of xylose and requires NAD for activity [1]. The overall efficiency of xylose conversion, as well as the endproducts of xylose metabolism, is connected through a complex regulatory network with the ability of XR and XDH to provide a high flux of carbon through the initial pathway. Therefore, XR and XDH are important targets for metabolic engineering of yeasts towards utilization of xylose in a manner that meets the requirements of biotech industry [1,2]. We have recently isolated XDH from the xyloseassimilating yeast Galactocandida mastotermitis (GmXDH) and performed a detailed characterization of the enzyme [3]. GmXDH is a zincdependent tetrameric polyol dehydrogenase, whose preferred substrates are xylitol (3600 M 1 s 1) and D-sorbitol (2000 M 1 s 1). At pH 7.5, the kinetic mechanism of D-sorbitol oxidation appeared to be steady-state ordered with cosubstrate binding first and leaving last. Dissociation of NADH and NAD seemed to be mainly rate-limiting in forward and reverse reactions, respectively. The gene encoding GmXDH has been cloned, sequenced and expressed in Escherichia coli [4]. Based on similarities at the level of primary structure, GmXDH has been classified as a member of the medium-chain dehydrogenase/ reductase (MDR) superfamily [5] and shows closest relationship with eukaryotic sorbitol/xylitol dehydrogenases. The tetrad of ligands to the catalytic zinc in mammalian sorbitol dehydrogenase, Cys-44/His-69/Glu-70/Glu-153 (numbering of the rat enzyme) [6 /8], is positionally conserved in GmXDH. The structure of the rat sorbitol dehydrogenase /NAD complex has recently ˚ resolution [9] and revealed been determined at 3 A a tetrahedral coordination of zinc by four protein ligands rather than three protein ligands and water which is the classical coordination of zinc in horse liver alcohol dehydrogenase [10]. We would like to have a better understanding of the kinetic and chemical mechanisms of XDH and are interested in how the enzyme achieves its high specificity for substrate and co-substrate. We believe that such information would be of general
interest in the context of MDR structure-function relationships and important for a successful engineering of the catabolic pathway of xylose in a heterologous environment, such as Saccharomyces cerevisiae [2]. This paper reports on the isolation and characterization of recombinant GmXDH expressed in E. coli . Structural and functional properties of the recombinant enzyme are compared with those of natural GmXDH. A detailed kinetic analysis has been carried out for NAD dependent oxidations of xylitol and D-sorbitol and the corresponding reverse reactions using initialrate data and isotope effects. The paper provides a basic set of kinetic parameters for wild-type GmXDH and knowledge of the stereochemistry of hydride transfer from NADH.
2. Materials and methods 2.1. Enzyme production and purification Production of GmXDH in E. coli JM 109 used the plasmid expression vector pBTac1 [4] and media reported recently [11]. Cultivation was carried out at 37 8C in shaken flasks with agitation at 110 rpm. Recombinant protein production was induced when the culture had reached an optical density of :/1.8. Before the addition of the inductor isopropyl-b-D-thio-galactopyranoside (0.45 mM), the incubation temperature was decreased to 25 8C to achieve an optimum yield of soluble protein. Work-up of bacterial cells and purification of GmXDH by dye ligand column chromatography using Procion Red HE3B (Red 120) immobilized onto Sepharose 4B-CL were carried out essentially as described elsewhere in more detail [3]. Standard assays for enzyme activity were conducted in the direction of xylitol oxidation at pH 10.0 and 25 8C by measuring the reduction of NAD at 340 nm [3]. Enzyme activity (1 U) corresponds to 1 mmol NADH produced per minute under these conditions. 2.2. Production of deuterium-labeled substrates Deuterium-labeling in the position of the 4-proR or 4-pro-S hydrogen on the nicotinamide ring
B. Nidetzky et al. / Chemico-Biological Interactions 143 /144 (2003) 533 /542
of NADH was carried out by reported enzymatic protocols [12]. 2-[2H]-D-Sorbitol was prepared enzymatically by a procedure analogous to that for 2-[2H]-D-mannitol synthesis described recently [13], except that GmXDH was used to stereospecifically reduce D-fructose and D-xylulose, respectively. Unlabeled D-sorbitol was prepared in exactly the same way as the deuterated compound, but using H-CO2 instead of [2H]-CO2 as donor of reducing equivalents. D-Xylulose was synthesized by microbial oxidation of D-arabinitol [14]. Judging from HPLC and TLC analyses in which authentic material was used as standard, 2-[1H] and 2-[2H]-D-sorbitol contained :/10 to 15% unreacted D-fructose, which was not removed during the ion exchange purification steps. Therefore, special care (pH control, T /55 8C) was taken to avoid browning reactions during concentration of synthesized D-sorbitol by evaporation. In all compounds, the correct position of the deuterium label was verified by standard NMR methods using a Bruker Avance 300 spectrometer. The degree of deuteration at the respective position was estimated to be /98%. 2.3. Characterization of recombinant GmXDH Zinc and magnesium contents of purified enzyme were determined by inductively coupled plasma mass spectrometry as described recently [3]. The molecular mass of functional GmXDH was determined by gel filtration using a Superose 12 HR 10/30 column equilibrated with 20 mM potassium phosphate, pH 7.1, containing 1 M NaCl. The enzyme was diluted into the same buffer to give a final concentration of 0.5 mg ml 1 and 0.2 ml of the protein solution was loaded onto the column. Elution was carried out ¨ ktaexplorer at a flow rate of 0.5 ml min 1 using A 100 for flow control and absorbance detection at 280 nm. Molecular mass calibration of the column was carried out using bovine serum albumin (67 kDa), aldolase (158 kDa), catalase (232 kDa) and ferritin (440 kDa). The stereospecificity of hydrogen transfer from NADH catalyzed by GmXDH was determined by a published protocol using 1HNMR [12]. Briefly, 4-R [2H]-NADH or 4-S [2H]NADH (:/3 mg each) were incubated in the
535
presence of a large, :/100-fold molar excess of D-fructose and 30 U GmXDH. The reactions were carried out at 25 8C in D2O using a 50 mM potassium phosphate buffer, pD 7.0. The decrease of NADH was monitored spectrophotometrically at 340 nm over time. When all NADH present had been oxidized, the samples were used for 1H-NMR analysis without further treatment. 2.4. Initial rate measurements and determination of kinetic parameters Initial rates of polyol oxidation and ketose reduction were measured at 259/0.5 8C by recording with a Beckman DU-650 spectrophotometer, the increase and decrease in the absorbance of NADH at 340 nm, respectively. We used a 50 mM potassium phosphate buffer, pH 7.5, or a 50 mM Tris /HCl buffer, pH 7.5. In all assays, the enzyme concentration was chosen such that a plot of absorbance against the reaction time was linear for 1 min, at least. If not mentioned otherwise, initial-rate data were acquired under conditions in which the substrate (or the coenzyme) was varied and a constant and saturating concentration of the coenzyme (or the substrate) was present. Kinetic parameters were obtained from unweighted nonlinear least-square fits of Eq. (1) to initial-rate data for which graphical analysis had shown that they yield linear double reciprocal plots. The Sigmaplot 2000 program (SPSS Inc.) was used and in Eq. (1), v is the initial rate, [E ] is the molar concentration of the enzyme subunit (37.4 kDa), [A ] is the substrate concentration, kcat is the turnover number (s 1) and KA is an apparent Michaelis constant. v kcat [E][A]=(KA [A])
(1)
Primary deuterium kinetic isotope effects (KIE) on apparent kinetic parameters for polyol oxidation and ketose reduction were obtained by directly comparing initial rates recorded with unlabeled and deuterium-labeled substrate and unlabeled and deuterium-labeled coenzyme, respectively. A 50 mM potassium phosphate buffer, pH 7.5, was used. KIEs were calculated by fitting Eq. (2) to the data where EV/K and EV are the isotope effects /1 on kcat/KA and kcat, respec-
536
B. Nidetzky et al. / Chemico-Biological Interactions 143 /144 (2003) 533 /542
tively. Fi is the fraction of deuterium in the coenzyme on the substrate. v kcat [E][A]=fKA (1Fi EV=K ) [A](1Fi EV )g
(2)
KIEs are reported using the nomenclature of Northrop [15] whereby Dkcat and Dkcat/K are primary deuterium KIEs on kcat and kcat/K , respectively.
3. Results and discussion 3.1. Production and purification of GmXDH Induced cell extracts of E. coli JM 109 contained :/380 U of xylitol dehydrogenase activity per milliliter. The specific enzyme activity based on the unit mass of protein was 15 U mg 1. Compared to wild-type strain of G. mastotermitis [3], recombinant enzyme production in E. coli led to increases in specific activity and total activity in the cell extract by factors of 8 and 35, respectively. The recombinant system eliminates the necessity to conduct pH- and dissolved oxygen-controlled bioreactor cultivations for the production of milligram amounts of protein. GmXDH was purified by group-specific dye ligand affinity chromatography using a two-step procedure summarized in Table 1. Effectively complete adsorption of GmXDH onto the Red 120 dye column, such that significant losses of active enzyme to the wash-through fraction were avoided, required that
the conductivity of the cell extract be smaller than the conductivity of the equilibration buffer, 50 mM potassium phosphate buffer, pH 7.0. Approximately 1/2 mg of appropriately diluted protein was applied to each milliliter of affinity gel. For protein elution, a salt gradient in three steps (no NaCl, 0.2 M NaCl, 1 M NaCl) was employed using 50 mM potassium phosphate buffer, pH 7.2. (Note that the small increase in pH for the elution buffer compared to the equilibration buffer increases the selectivity of salt-induced protein elution.) Under the conditions used, GmXDH eluted at 0.2 M NaCl. The reversible adsorption behavior of GmXDH during NaCl-mediated elution chromatography on Red 120-Sepharose 4B-CL was not distinguishable from that of the natural enzyme reported by Lunzer et al. [3]. Re-chromatography of partially purified GmXDH under otherwise identical conditions just described yielded the pure enzyme, as demonstrated in Fig. 1. Natural GmXDH has previously been purified by using Red-120 biomimetic chromatography and affinity elution with NADH [3]. No attempts were made to use a similar procedure with the recombinant protein. NADH must be removed from holo GmXDH by dialysis or repeated gel filtration, both of which are factors of irreversible inactivation of the enzyme, likely because the active-site zinc is gradually released into bulk solution [3]. The specific activity of the isolated recombinant enzyme matched the published specific activity of purified natural GmXDH. Apparent kinetic para-
Table 1 Purification of GmXDH expressed in E. coli
Cell extract Red 120 /step 1b Red 120 /step 2b a
Volumetric activity (U ml 1)a
Specific activity (U mg 1)
Yield (%)
380 346 267
15 96 125
100 91 70
Data are based on processing 1 ml of cell extract at room temperature. An Amersham Pharmacia Biotech XK 26 column containing 23 ml of Red 120 Sepharose 4B-CL was used; :/1 /2 mg of protein were applied to each milliliter of gel taking care that the conductivity of the sample was below the conductivity of the 50 mM potassium phosphate equilibration buffer (pH 7.0); elution was carried out with a three-step gradient at 0, 0.2 and 1 M NaCl in 50 mM potassium phosphate buffer, pH 7.2; gel filtration with Sephadex G 25 was used to remove salt prior to the re-chromatography on the dye ligand column and concentration of protein solutions was carried out by ultrafiltration with an Amicon stirred cell or Filtron centrifugal microconcentrator tubes and a molecular mass cut-off of 10 kDa. b
B. Nidetzky et al. / Chemico-Biological Interactions 143 /144 (2003) 533 /542
Fig. 1. Purification of GmXDH documented by SDS PAGE. Visualization of protein bands was by staining with Coomassie brilliant blue. Lanes (left to right), 1: molecular mass standards (in kDa); 2: cell extract; Red 120 chromatography; Red 120 rechromatography.
meters of GmXDH for NAD -dependent reactions with D-sorbitol and xylitol at pH 10.0 and NADH-dependent reactions with D-fructose and D-xylulose at pH 7.0 were identical to corresponding kinetic parameters measured with the natural enzyme (not shown). 3.2. Structural and functional properties of GmXDH The molecular mass of GmXDH subunit was determined by SDS-PAGE (Fig. 1) and is 38 (9/1) kDa. This is the expected size for the full-length protomer of xylitol dehydrogenase. In sizing chromatography using a Superose 12 HR 10/30 gelfiltration column, GmXDH eluted as a single protein peak with an apparent molecular mass of 160 kDa. Hence, similar to natural xylitol dehydrogenase, the recombinant enzyme is a functional tetramer and appears not to exist as a mixture of different oligomeric forms in solution. The pI of GmXDH is :/6.0, in good agreement with the pI value of the natural enzyme [3]. GmXDH contained 0.99/0.2 mole of Zn2 per each mole of 37.4-kDa protomer, indicating that zinc stoichiometry of the natural enzyme had been preserved completely during recombinant protein production. In contrast to the natural enzyme that was
537
found to contain 6 Mg2 per protein subunit, GmXDH preparations did not show a significant content of Mg2. We cannot offer an explanation for this difference on the basis of data available at the present time. The heterologous environment in E. coli as well as differences in protocols used for purification of natural and recombinant xylitol dehydrogenase could be relevant factors. The stereospecificity of hydrogen transfer to and from NADH was determined by 1H-NMR, essentially as described recently for Candida tenuis xylose reductase [12]. When 4-R [2H]-NADH was employed as co-substrate for reduction of Dfructose by GmXDH, the deuterium was depleted such that after complete oxidation of labeled NADH authentic NAD was obtained (not shown). By contrast, when 4-S [2H]-NADH was used, the hydrogen was depleted at the C4 position of the nicotinamide ring. The product revealed clearly by NMR analysis was [4-2H]-NAD . Therefore, GmXDH is specific for transferring the 4-pro-R hydrogen of NADH. The observed (‘A-side’) stereospecificity of hydrogen transfer is typical of members of the MDR superfamily [16]. 3.3. Kinetics Initial rates of NAD -dependent oxidation of xylitol and D-sorbitol (forward reaction) and NADH-dependent reduction of D-xylulose and D-fructose (reverse reaction) were recorded in the absence of products at pH 7.5 and 25 8C. In each direction of the reaction, they were acquired under conditions in which the substrate was varied at several constant levels of the coenzyme (Fig. 2a, b). Graphical analysis of the data for polyol oxidations in double reciprocal plots of the form 1/v versus 1/[polyol] gave intersecting patterns with crossover points that were to the left of the vertical axis and above the horizontal axis (Fig. 2a). The same analysis for initial rates of Dfructose reduction yielded a crossover point as far below the horizontal axis as it was above the same axis in the direction of D-sorbitol oxidation (not shown). Note that this is a requirement of the Theorell /Chance mechanism although it does not, of course, prove the mechanism [17]. The result is an indication, however, that the mechanism is not
538
B. Nidetzky et al. / Chemico-Biological Interactions 143 /144 (2003) 533 /542
Fig. 2. Double reciprocal plots of initial rates of NAD -dependent oxidation of xylitol (a) and NADH-dependent reduction of Dxylulose (b) by GmXDH measured at varied substrate concentrations and at several constant levels of co-substrate, as indicated in the plots. In the left-hand side panel (a), NAD concentrations were 0.1 (full squares), 0.2 (open triangles), 0.5 (full triangles), 1.0 (open circles) and 2.0 mM (full circles). In the right-hand side panel (b), NADH concentrations were 20 (open triangles), 50 (full triangles), 100 (open circles) and 200 mM (full circles). Data are for reactions in 50 mM potassium phosphate buffer, pH 7.5. The symbols show the mean values of two independent determinations and lines represent the fit of the shown data to Eq. (3).
rapid equilibrium random. Results for D-xylulose reduction gave initial rate patterns that were intersecting (Fig. 2b). However, the crossover point was not as clearly defined as in the opposite direction of reaction. According to the fit of the data to Eq. (3), it lay above the horizontal axis, as shown in Fig. 2(b). Cleland has discussed that in a sequential bi-reactant mechanism, where both substrates dissociate more slowly than they react to yield products, initial rate patterns may occur that are not clearly intersecting and may occasionally look nearly parallel [17]. Isotope effect data (not shown) suggest that both D-xylulose and NADH may be sticky indeed. The experimental data in Fig. 2 were fitted to Eq. (3) and with the assumption of ordered reactant binding, [A ] and [B ] are coenzyme and substrate, respectively, KiA is the apparent dissociation constant of enzyme / coenzyme complex and KA and KB are Michaelis constants.
v kcat [E][A][B]= (KiA KB KB [A]KA [B][A][B])
(3)
Estimates for the kinetic parameters are summarized in Table 2. The internal consistency of each set of kinetic parameters was checked with the Haldane relationship for a sequential bi-bi kinetic mechanism. The equilibrium constant (Keq) calculated by using kinetic parameters for reactions with xylitol and D-xylulose in Tris buffer was 5.1 /10 11 M. This value for Keq is in good agreement with the experimental equilibrium constant of 7 /10 11 M for the reaction, xylitol/ NAD l/D-xylulose/NADH/H [18]. Likewise, the calculated equilibrium constant of 2.9 /10 9 M for the enzymic reactions with Dsorbitol and D-fructose agrees with the value of 3.7 /10 9 M determined experimentally [19]. Keq for reactions with xylitol and D-xylulose in phos-
B. Nidetzky et al. / Chemico-Biological Interactions 143 /144 (2003) 533 /542
539
Table 2 Kinetic parameters of GmXDH at 25 8C and pH 7.5 Parameter
50 mM Tris /HCl buffer a D-Sorbitol
kcat (s 1) KiNAD (mM) KNAD (mM) Kpolyol (mM)
kcat (s 1) KiNADH (mM) KNADH (mM) Kketose (mM)
1149/2 7409/110 2279/18 559/4
50 mM Potassium phosphate buffer Xylitol
Xylitol
1439/4 4109/100 3209/32 399/4
1709/8 7809/194 1409/20 129/2
a D-Fructose
D-Xylulose
D-Xylulose
9209/120 239/9 (249/5)b 1609/30 13009/200
15009/200 279/6 869/8 109/1
18009/350 (1909/110)c 739/24 89/4
5.1 /10 11
4.8 /10 10
Haldane Keq (M)
2.9 /10 9
a Analysis of the crossover point for intersecting initial rate pattern can be carried out by using the relationship (1/v )X over / (kcat[E ]) 1 (1 /KNAD(H)/KiNAD(H)) [17] where (1/v )X over is the crossover point. b Determined from linear competitive inhibition of NADH against NAD . c This value for KiNADH should not be interpreted as an apparent dissociation constant of enzyme /NADH complex.
phate buffer was 4.8 /10 10 M and about seven times the value expected. Interpretation of the KiNAD(H) values in Table 2 as apparent dissociation constants for NADH and NAD requires that the kinetic mechanism be ordered with co-substrate binding first and leaving last. A value of 0.8 mM for KiNAD has recently been calculated from initial-rate data for NAD dependent oxidation of D-sorbitol in phosphate buffer by GmXDH [3]. It is in good agreement with the KiNAD values found here in both phosphate- and Tris-buffered solutions (Table 2), suggesting that KiNAD could reflect binding of NAD to the free enzyme. An estimate of 10 mM for KiNADH has been obtained from steady-state kinetic analysis of D-fructose reduction in phosphate buffer by GmXDH [3]. Equilibrium binding studies, in which NADH was used as inhibitor against co-substrate in the NAD -dependent oxidation of D-sorbitol, yielded a closely similar value of 12 mM for KiNADH [3]. Likewise, three independent estimates for KiNADH have been obtained herein using Tris-buffered solution. Though about 2.5 times higher than the corresponding KiNADH values measured in phosphate buffer, the values for KiNADH in Table 2 were
nearly identical (:/25 mM), as expected if NADH bound to the free enzyme under the conditions used. It is interesting, therefore, that steady-state kinetic analysis of D-xylulose reduction in phosphate buffer yielded a value of 190 mM for KiNADH, which is almost 20 times that determined in the previous study [3]. Although the value is illdefined statistically, it is very probably /25 mM. This result shows that KiNADH does not represent an apparent dissociation constant of the enzyme / NADH complex and might imply a random mechanism of reaction of GmXDH with D-xylulose and NADH. If binding of D-xylulose and NADH to GmXDH were completely random and rapiD-equilibrium conditions applied, the relationship KA/KiA /a could be used to calculate a value of 0.39 for a , which is the factor by which binding of one substrate changes the dissociation constant for the other substrate. The fact that a B/1 implies that the affinity for D-xylulose increases when NADH is bound and likewise the affinity for NADH increases when D-xylulose is bound. Synergism between co-substrate and substrate binding has been also observed with liver alcohol dehydrogenase [20,21].
540
B. Nidetzky et al. / Chemico-Biological Interactions 143 /144 (2003) 533 /542
Comparisons of kcat values for reductions of Dxylulose and D-fructose are interesting (Table 2 and Ref. [3]). Clearly, kcat for reduction of Dxylulose is :/1.6- to 2.3-fold higher than kcat for reduction of D-fructose. It has been shown recently that kcat for reduction of D-fructose in phosphate buffer, pH 7.5, is (completely) limited by dissociation of the enzyme /NAD complex. Release of NAD takes place with an estimated rate constant of 779 s 1 [3]. Now, if reaction of GmXDH in the presence of saturating concentrations of Dxylulose went through a compulsorily ordered mechanism involving enzyme /NAD complex, turnover number could not have a value larger than :/800 s 1, at variance with the experimental observations. The results imply that dissociation of NAD cannot be fully rate-limiting for turnover during D-xylulose reduction. A plausible scenario would be that an abortive enzyme / NAD /D-xylulose complex is formed at high Dxylulose levels which releases NAD between 1.6and 2.3-fold more rapidly than the enzyme / NAD complex. In that case, the binary complex with D-xylulose could be important kinetically and contribute to the distribution of the enzyme in the steady state. Protection afforded by 20 mM Dxylulose against inactivation of GmXDH by metal chelators such as EDTA suggests that D-xylulose is capable of binding to the free enzyme (not shown). In the literature on MDRs, there is precedence for kinetic mechanisms involving abortive complexes Table 3 Kinetic isotope effects on kinetic parameters for NADHdependent reduction of D-fructose and NAD -dependent oxidation of D-sorbitol of GmXDH, determined in 50 mM potassium phosphate buffer, pH 7.5, and at 25 8C Isotope effect
Reduction
D
n.a.a 1.829/0.25 1.209/0.16
kcat kcat/Kfructose D kcat/KNADH D
Oxidation D
kcat kcat/Ksorbitol D kcat/KNAD D
a
1.109/0.10 1.939/0.20 1.209/0.21
n.a., Not applicable because saturation in substrate could not be achieved.
and kinetically significant enzyme /substrate complexes [22 /25]. For example, an enzyme /NADH / alcohol complex is probably physiologically significant in reaction catalyzed by human liver alcohol dehydrogenase g2 [24]. In an alternative interpretation of our data, however, release of NAD could occur from the enzyme /xylitol / NAD complex, implying a random component to the mechanism. Clearly, more data will be required to establish a complete kinetic mechanism of GmXDH for the physiological reaction, xylitol/NAD l/D-xylulose/NADH/H . 3.4. Kinetic isotope effects Isotope effects on kinetic parameters for NADH-dependent reduction of D-fructose were measured by comparing initial rates recorded in 50 mM phosphate buffer, pH 7.5, and at 25 8C in the presence of unlabeled co-substrate and 4-R [2H]NADH. Data were collected under conditions in which either co-substrate was constant at 220 mM (:/3 /KNADH) or substrate was constant at a level of 1.5 M (:/KB for D-fructose). Data were fitted to Eq. (2) and results are shown in Table 3. Kinetic isotope effects on kinetic parameters for oxidation of D-sorbitol were determined under the above described buffer and temperature conditions from a comparison of initial rates recorded in the presence of unlabeled substrate and 2-[2H]-labeled polyol. Data measured at constant co-substrate concentration (3 mM; /5 /KNAD) or constant substrate concentration (300 mM; /5 /Kpolyol) were fitted to Eq. (2) and isotope effects are shown in Table 3. In the direction of D-fructose reduction, Dkcat/ Kfructose /Dkcat/KNADH and therefore, the release of D-fructose from the central complex must take place at a rate faster than the rate of release of NADH. The value for Dkcat/KNADH contains uncertainty because saturation in D-fructose could not be achieved in the experiment (Kfructose /1 M). (An isotope effect on kcat is, therefore, not available.) If the mechanism of GmXDH were strictly ordered and NADH was the substrate binding first, no isotope effect would be expected on kcat/KNADH in the presence of saturating concentrations of the second substrate. In that
B. Nidetzky et al. / Chemico-Biological Interactions 143 /144 (2003) 533 /542
case, the commitment of enzyme /NADH to undergo reaction as opposed to dissociate would become infinite and thus the isotope effect on the hydride transfer step would be completely masked [26]. The level of the second substrate will affect the magnitude of the commitment to catalysis and the concentration of D-fructose required to halfsuppress the isotope effect on kcat/KNADH would be given by the expression KiNADHKfructose/ KNADH. Taking data from Lunzer et al. [3], we calculate this concentration to be 0.32 M, whereas the actual concentration used in the experiment was 1.5 M. Considering that the value of Dkcat/ KNADH is actually close to and perhaps statistically not different from 1, the results agree with a mechanism of NADH-dependent reduction of Dfructose, in which reactant binding is (preferentially) ordered. The isotope effect on kcat for D-sorbitol oxidation has a value very close to 1.00, suggesting that product dissociation is completely rate-limiting for the overall reaction at pH 7.5. As shown in Table 3, Dkcat/Ksorbitol /Dkcat/KNAD and Dkcat/KNAD has a small value of 1.2 that perhaps is not statistically different from 1.0. This result indicates that although NAD might be released from the ternary complex at a finite, however small rate, the rate of dissociation of NAD is clearly very much slower than the corresponding dissociation rate of D-sorbitol. In conclusion, the isotope effect data for D-sorbitol oxidation and D-fructose reduction by GmXDH at pH 7.5 suggest an ordered kinetic mechanism of NAD -dependent D-sorbitol oxidation in which NAD binds before D-sorbitol and NADH is released only after Dfructose has dissociated. The rate of last product release is rate-limiting for kcat in this direction. The results, therefore, provide evidence supporting the Theorell /Chance mechanism proposed recently [3].
Acknowledgements Financial support from the Austrian Science Funds is gratefully acknowledged (FWF projects P-12569-MOB and P-15208-MOB to B.N.)
541
References [1] B. Hahn-Ha¨gerdal, H. Jeppson, K. Skoog, B.A. Prior, Biochemistry and physiology of xylose fermentation by yeasts, Enzyme Microb. Technol. 16 (1994) 933 /943. [2] B. Hahn-Ha¨gerdal, C.F. Wahlbom, M. Gardonyi, W.H. van Zyl, R.R. Cordero Otero, L.J. Jonsson, Metabolic engineering of Saccharomyces cerevisiae for xylose utilization, Adv. Biochem. Eng. Biotechnol. 73 (2001) 53 /84. [3] R. Lunzer, Y. Mamnun, D. Haltrich, K.D. Kulbe, B. Nidetzky, Structural and functional properties of a yeast xylitol dehydrogenase, a Zn2 -containing metalloenzyme similar to medium-chain sorbitol dehydrogenases, Biochem. J. 336 (1998) 91 /99. [4] A. Habenicht, H. Motejadded, M. Kiess, A. Wegerer, R. Mattes, Xylose utilisation: cloning and characterisation of the xylitol dehydrogenase from Galactocandida mastotermitis , Biol. Chem. 380 (1999) 1405 /1411. [5] B. Persson, J.S. Zigler, H. Jo¨rnvall, A super-family of medium-chain dehydrogenases/reductases (MDR). Sublines including z-crystallin, alcohol and polyol dehydrogenases, quinone oxidoreductases, enoyl reductases, VAT-1 and other proteins, Eur. J. Biochem. 226 (1994) 15 /22. [6] C. Karlsson, J.-O. Ho¨o¨g, Zinc coordination in mammalian sorbitol dehydrogenase, Eur. J. Biochem. 216 (1993) 103 / 107. [7] J. Jeffery, H. Jo¨rnvall, Sorbitol dehydrogenase, Adv. Enzymol. 61 (1988) 47 /106. [8] M.J. Banfield, M.E. Salvucci, E.N. Baker, C.A. Smith, Crystal structure of the NADP(H)-dependent ketose ˚ resolution, J. reductase from Bemisia argentifolii at 2.3 A Mol. Biol. 306 (2001) 239 /250. [9] K. Johansson, M. El-Ahmad, C. Kaiser, H. Jo¨rnvall, H. Eklund, J.-O. Ho¨o¨g, S. Ramaswamy, Crystal structure of sorbitol dehydrogenase, Chem.-Biol. Interact. 130 /132 (2001) 351 /358. [10] H. Eklund, E. Horjales, H. Jo¨rnvall, C.-I. Bra¨nde´n, J. Jeffery, Molecular aspects of functional differences between alcohol and sorbitol dehydrogenases, Biochemistry 24 (1985) 8005 /8012. [11] P. Mayr, K. Bru¨ggler, K.D. Kulbe, B. Nidetzky, Xylose metabolism in Candida intermedia : efficient isolation and characterisation of two aldose reductases with different coenzyme specificities, J. Chromatogr. B 737 (2000) 195 / 202. [12] B. Nidetzky, P. Mayr, W. Neuhauser, M. Puchberger, Structural and functional properties of aldose (xylose) reductase from the D-xylose-metabolizing yeast Candida tenuis , Chem.-Biol. Interact. 130 /132 (2001) 583 /595. [13] M. Klimacek, B. Nidetzky, Examining the relative timing of hydrogen abstraction steps during NAD -dependent oxidation of secondary alcohols catalyzed by long-chain mannitol dehydrogenase from Pseudomonas fluorescens using pH and isotope effects, Biochemistry 41 (2002) 10158 /10165.
542
B. Nidetzky et al. / Chemico-Biological Interactions 143 /144 (2003) 533 /542
[14] G. Mayer, K.D. Kulbe, B. Nidetzky, Utilization of xylitol dehydrogenase in a combined microbial/enzymatic process for production of xylitol from D-glucose, Appl. Biochem. Biotechnol. 98 /100 (2002) 577 /589. [15] D.B. Northrop, Deuterium and tritium kinetic isotope effects on initial rates, Methods Enzymol. 87 (1982) 607 / 625. [16] R. Meijers, R.J. Morris, H.W. Adolph, A. Merli, V.S. Lamzin, E.S. Cedergren-Zeppezauer, On the enzymatic activation of NADH, J. Biol. Chem. 276 (2001) 9316 / 9321. [17] W.W. Cleland, Determining the chemical mechanisms of enzyme-catalyzed reactions by kinetic studies, Adv. Enzymol. 45 (1977) 273 /387. [18] M. Rizzi, K. Harwart, N.-A. Bui-Thanh, H. Dellweg, A kinetic study of the NAD -xylitol dehydrogenase from the yeast Pichia stipitis , J. Ferment. Bioeng. 67 (1989) 25 / 30. [19] R.I. Lindstad, L.F. Hermansen, J. McKinley-McKee, The kinetic mechanism of sheep liver sorbitol dehydrogenase, Eur. J. Biochem. 210 (1992) 641 /647. [20] P. Andersson, J. Kvassman, B. Olde´n, G. Pettersson, Synergism between coenzyme and aldehyde binding to
[21]
[22]
[23]
[24]
[25]
[26]
liver alcohol dehydrogenase, Eur. J. Biochem. 133 (1983) 651 /655. P. Andersson, J. Kvassman, B. Olde´n, G. Pettersson, Synergism between coenzyme and alcohol binding to liver alcohol dehydrogenase, Eur. J. Biochem. 144 (1984) 317 / 324. P. Andersson, J. Kvassman, B. Olde´n, G. Pettersson, Catalytic significance of binary enzyme /aldehyde complexes in the liver alcohol dehydrogenase reaction, Eur. J. Biochem. 139 (1984) 519 /527. F.M. Dickinson, G.P. Monger, A study of the kinetics and mechanism of yeast alcohol dehydrogenase with a variety of substrates, Biochem. J. 131 (1973) 261 /270. H.A. Charlier, Jr, B.V. Plapp, Kinetic cooperativity of human liver alcohol dehydrogenase g2, J. Biol. Chem. 275 (2000) 11569 /11575. K. Dalziel, F.M. Dickinson, Substrate activation and inhibition in coenzyme /substrate reactions, Biochem. J. 100 (1966) 491 /500. P.F. Cook, Kinetic and regulatory mechanisms of enzymes from isotope effects, in: P.F. Cook (Ed.), Enzyme Mechanisms From Isotope Effects, CRC Press, Boca Raton, 1991, p. 203.
Characterization of recombinant xylitol dehydrogenase from Galactocandida mastotermitis expressed in Escherichia coli Bernd Nidetzky a,*, Heidemarie Helmer a,b, Mario Klimacek a, Regina Lunzer a,b, Gerhard Mayer a,b b
a Institute of Biotechnology, Graz University of Technology, Petersgasse 12/I, A-8010 Graz, Austria Institute of Food Technology, University of Agricultural Sciences Vienna, Muthgasse 18, A-1190 Vienna, Austria
Abstract The plasmid-encoded gene of xylitol dehydrogenase from the yeast Galactocandida mastotermitis was expressed in Escherichia coli at 25 8C. Recombinant enzyme was isolated in 70% yield using two steps of biomimetic affinity chromatography with the dye ligand Procion Red HE3B immobilized onto Sepharose 4B-CL. Similar to natural enzyme, recombinant xylitol dehydrogenase is a functional homotetramer with a stoichiometric content of catalytic zinc in each 37-kDa subunit. Though lacking bound Mg2 found in xylitol dehydrogenase isolated from yeast cell extracts, the recombinant enzyme is as active and stable as the native enzyme. Stereospecificity of enzymic hydrogen transfer from NADH has been determined by 1H-NMR and is 4-pro-R . A detailed steady-state kinetic analysis has been carried out for the enzymic reaction, polyol/NAD l/ketose/NADH/H , at pH 7.5 and 25 8C using xylitol and Dxylulose, the physiological polyol /ketose pair, as well as D-sorbitol and D-fructose. Primary deuterium kinetic isotope effects on steady-state kinetic parameters for oxidation of D-sorbitol and reduction of D-fructose have been measured at pH 7.5. Combined results of initial-rate analysis and isotope effect studies suggest that the enzyme utilizes a preferentially ordered kinetic mechanism in which NAD binds before D-sorbitol and D-fructose is released before NADH. Dissociation of NADH appears to be the main rate-limiting step for D-sorbitol oxidation under substratesaturated reaction conditions. # 2002 Elsevier Science Ireland Ltd. All rights reserved. Keywords: Xylose metabolism; Medium-chain dehydrogenase/reductase; Kinetic mechanism; Stereospecificity
1. Introduction
Abbreviations: GmXDH, recombinant XDH from Galactocandida mastotermitis expressed in Escherichia coli ; KIE, primary deuterium kinetic isotope effect; MDR, medium-chain dehydrogenase/reductase; XDH, xylitol dehydrogenase; XR, xylose reductase. * Corresponding author. Tel.: /43-316-873-8400; fax: /43316-873-8434. E-mail address: [email protected] (B. Nidetzky).
Xylose reductase (alditol:NAD(P) -1-oxidoreductase; XR; EC 1.1.1.21) and xylitol dehydrogenase (xylitol:NAD -2-oxidoreductase; XDH; EC 1.1.1.9) constitute the initial catabolic pathway for D-xylose in yeasts and fungi. The net reaction of this pathway is an isomerization of D-xylose into D-xylulose which is then phosphorylated at the 5-OH and metabolized further via pentose
0009-2797/02/$ - see front matter # 2002 Elsevier Science Ireland Ltd. All rights reserved. PII: S 0 0 0 9 - 2 7 9 7 ( 0 2 ) 0 0 2 1 5 - 6
534
B. Nidetzky et al. / Chemico-Biological Interactions 143 /144 (2003) 533 /542
phosphate and Embden /Meyerhof /Parnas pathways [1]. XDH is responsible for the oxidation of xylitol, resulting from the XR-catalyzed reduction of xylose and requires NAD for activity [1]. The overall efficiency of xylose conversion, as well as the endproducts of xylose metabolism, is connected through a complex regulatory network with the ability of XR and XDH to provide a high flux of carbon through the initial pathway. Therefore, XR and XDH are important targets for metabolic engineering of yeasts towards utilization of xylose in a manner that meets the requirements of biotech industry [1,2]. We have recently isolated XDH from the xyloseassimilating yeast Galactocandida mastotermitis (GmXDH) and performed a detailed characterization of the enzyme [3]. GmXDH is a zincdependent tetrameric polyol dehydrogenase, whose preferred substrates are xylitol (3600 M 1 s 1) and D-sorbitol (2000 M 1 s 1). At pH 7.5, the kinetic mechanism of D-sorbitol oxidation appeared to be steady-state ordered with cosubstrate binding first and leaving last. Dissociation of NADH and NAD seemed to be mainly rate-limiting in forward and reverse reactions, respectively. The gene encoding GmXDH has been cloned, sequenced and expressed in Escherichia coli [4]. Based on similarities at the level of primary structure, GmXDH has been classified as a member of the medium-chain dehydrogenase/ reductase (MDR) superfamily [5] and shows closest relationship with eukaryotic sorbitol/xylitol dehydrogenases. The tetrad of ligands to the catalytic zinc in mammalian sorbitol dehydrogenase, Cys-44/His-69/Glu-70/Glu-153 (numbering of the rat enzyme) [6 /8], is positionally conserved in GmXDH. The structure of the rat sorbitol dehydrogenase /NAD complex has recently ˚ resolution [9] and revealed been determined at 3 A a tetrahedral coordination of zinc by four protein ligands rather than three protein ligands and water which is the classical coordination of zinc in horse liver alcohol dehydrogenase [10]. We would like to have a better understanding of the kinetic and chemical mechanisms of XDH and are interested in how the enzyme achieves its high specificity for substrate and co-substrate. We believe that such information would be of general
interest in the context of MDR structure-function relationships and important for a successful engineering of the catabolic pathway of xylose in a heterologous environment, such as Saccharomyces cerevisiae [2]. This paper reports on the isolation and characterization of recombinant GmXDH expressed in E. coli . Structural and functional properties of the recombinant enzyme are compared with those of natural GmXDH. A detailed kinetic analysis has been carried out for NAD dependent oxidations of xylitol and D-sorbitol and the corresponding reverse reactions using initialrate data and isotope effects. The paper provides a basic set of kinetic parameters for wild-type GmXDH and knowledge of the stereochemistry of hydride transfer from NADH.
2. Materials and methods 2.1. Enzyme production and purification Production of GmXDH in E. coli JM 109 used the plasmid expression vector pBTac1 [4] and media reported recently [11]. Cultivation was carried out at 37 8C in shaken flasks with agitation at 110 rpm. Recombinant protein production was induced when the culture had reached an optical density of :/1.8. Before the addition of the inductor isopropyl-b-D-thio-galactopyranoside (0.45 mM), the incubation temperature was decreased to 25 8C to achieve an optimum yield of soluble protein. Work-up of bacterial cells and purification of GmXDH by dye ligand column chromatography using Procion Red HE3B (Red 120) immobilized onto Sepharose 4B-CL were carried out essentially as described elsewhere in more detail [3]. Standard assays for enzyme activity were conducted in the direction of xylitol oxidation at pH 10.0 and 25 8C by measuring the reduction of NAD at 340 nm [3]. Enzyme activity (1 U) corresponds to 1 mmol NADH produced per minute under these conditions. 2.2. Production of deuterium-labeled substrates Deuterium-labeling in the position of the 4-proR or 4-pro-S hydrogen on the nicotinamide ring
B. Nidetzky et al. / Chemico-Biological Interactions 143 /144 (2003) 533 /542
of NADH was carried out by reported enzymatic protocols [12]. 2-[2H]-D-Sorbitol was prepared enzymatically by a procedure analogous to that for 2-[2H]-D-mannitol synthesis described recently [13], except that GmXDH was used to stereospecifically reduce D-fructose and D-xylulose, respectively. Unlabeled D-sorbitol was prepared in exactly the same way as the deuterated compound, but using H-CO2 instead of [2H]-CO2 as donor of reducing equivalents. D-Xylulose was synthesized by microbial oxidation of D-arabinitol [14]. Judging from HPLC and TLC analyses in which authentic material was used as standard, 2-[1H] and 2-[2H]-D-sorbitol contained :/10 to 15% unreacted D-fructose, which was not removed during the ion exchange purification steps. Therefore, special care (pH control, T /55 8C) was taken to avoid browning reactions during concentration of synthesized D-sorbitol by evaporation. In all compounds, the correct position of the deuterium label was verified by standard NMR methods using a Bruker Avance 300 spectrometer. The degree of deuteration at the respective position was estimated to be /98%. 2.3. Characterization of recombinant GmXDH Zinc and magnesium contents of purified enzyme were determined by inductively coupled plasma mass spectrometry as described recently [3]. The molecular mass of functional GmXDH was determined by gel filtration using a Superose 12 HR 10/30 column equilibrated with 20 mM potassium phosphate, pH 7.1, containing 1 M NaCl. The enzyme was diluted into the same buffer to give a final concentration of 0.5 mg ml 1 and 0.2 ml of the protein solution was loaded onto the column. Elution was carried out ¨ ktaexplorer at a flow rate of 0.5 ml min 1 using A 100 for flow control and absorbance detection at 280 nm. Molecular mass calibration of the column was carried out using bovine serum albumin (67 kDa), aldolase (158 kDa), catalase (232 kDa) and ferritin (440 kDa). The stereospecificity of hydrogen transfer from NADH catalyzed by GmXDH was determined by a published protocol using 1HNMR [12]. Briefly, 4-R [2H]-NADH or 4-S [2H]NADH (:/3 mg each) were incubated in the
535
presence of a large, :/100-fold molar excess of D-fructose and 30 U GmXDH. The reactions were carried out at 25 8C in D2O using a 50 mM potassium phosphate buffer, pD 7.0. The decrease of NADH was monitored spectrophotometrically at 340 nm over time. When all NADH present had been oxidized, the samples were used for 1H-NMR analysis without further treatment. 2.4. Initial rate measurements and determination of kinetic parameters Initial rates of polyol oxidation and ketose reduction were measured at 259/0.5 8C by recording with a Beckman DU-650 spectrophotometer, the increase and decrease in the absorbance of NADH at 340 nm, respectively. We used a 50 mM potassium phosphate buffer, pH 7.5, or a 50 mM Tris /HCl buffer, pH 7.5. In all assays, the enzyme concentration was chosen such that a plot of absorbance against the reaction time was linear for 1 min, at least. If not mentioned otherwise, initial-rate data were acquired under conditions in which the substrate (or the coenzyme) was varied and a constant and saturating concentration of the coenzyme (or the substrate) was present. Kinetic parameters were obtained from unweighted nonlinear least-square fits of Eq. (1) to initial-rate data for which graphical analysis had shown that they yield linear double reciprocal plots. The Sigmaplot 2000 program (SPSS Inc.) was used and in Eq. (1), v is the initial rate, [E ] is the molar concentration of the enzyme subunit (37.4 kDa), [A ] is the substrate concentration, kcat is the turnover number (s 1) and KA is an apparent Michaelis constant. v kcat [E][A]=(KA [A])
(1)
Primary deuterium kinetic isotope effects (KIE) on apparent kinetic parameters for polyol oxidation and ketose reduction were obtained by directly comparing initial rates recorded with unlabeled and deuterium-labeled substrate and unlabeled and deuterium-labeled coenzyme, respectively. A 50 mM potassium phosphate buffer, pH 7.5, was used. KIEs were calculated by fitting Eq. (2) to the data where EV/K and EV are the isotope effects /1 on kcat/KA and kcat, respec-
536
B. Nidetzky et al. / Chemico-Biological Interactions 143 /144 (2003) 533 /542
tively. Fi is the fraction of deuterium in the coenzyme on the substrate. v kcat [E][A]=fKA (1Fi EV=K ) [A](1Fi EV )g
(2)
KIEs are reported using the nomenclature of Northrop [15] whereby Dkcat and Dkcat/K are primary deuterium KIEs on kcat and kcat/K , respectively.
3. Results and discussion 3.1. Production and purification of GmXDH Induced cell extracts of E. coli JM 109 contained :/380 U of xylitol dehydrogenase activity per milliliter. The specific enzyme activity based on the unit mass of protein was 15 U mg 1. Compared to wild-type strain of G. mastotermitis [3], recombinant enzyme production in E. coli led to increases in specific activity and total activity in the cell extract by factors of 8 and 35, respectively. The recombinant system eliminates the necessity to conduct pH- and dissolved oxygen-controlled bioreactor cultivations for the production of milligram amounts of protein. GmXDH was purified by group-specific dye ligand affinity chromatography using a two-step procedure summarized in Table 1. Effectively complete adsorption of GmXDH onto the Red 120 dye column, such that significant losses of active enzyme to the wash-through fraction were avoided, required that
the conductivity of the cell extract be smaller than the conductivity of the equilibration buffer, 50 mM potassium phosphate buffer, pH 7.0. Approximately 1/2 mg of appropriately diluted protein was applied to each milliliter of affinity gel. For protein elution, a salt gradient in three steps (no NaCl, 0.2 M NaCl, 1 M NaCl) was employed using 50 mM potassium phosphate buffer, pH 7.2. (Note that the small increase in pH for the elution buffer compared to the equilibration buffer increases the selectivity of salt-induced protein elution.) Under the conditions used, GmXDH eluted at 0.2 M NaCl. The reversible adsorption behavior of GmXDH during NaCl-mediated elution chromatography on Red 120-Sepharose 4B-CL was not distinguishable from that of the natural enzyme reported by Lunzer et al. [3]. Re-chromatography of partially purified GmXDH under otherwise identical conditions just described yielded the pure enzyme, as demonstrated in Fig. 1. Natural GmXDH has previously been purified by using Red-120 biomimetic chromatography and affinity elution with NADH [3]. No attempts were made to use a similar procedure with the recombinant protein. NADH must be removed from holo GmXDH by dialysis or repeated gel filtration, both of which are factors of irreversible inactivation of the enzyme, likely because the active-site zinc is gradually released into bulk solution [3]. The specific activity of the isolated recombinant enzyme matched the published specific activity of purified natural GmXDH. Apparent kinetic para-
Table 1 Purification of GmXDH expressed in E. coli
Cell extract Red 120 /step 1b Red 120 /step 2b a
Volumetric activity (U ml 1)a
Specific activity (U mg 1)
Yield (%)
380 346 267
15 96 125
100 91 70
Data are based on processing 1 ml of cell extract at room temperature. An Amersham Pharmacia Biotech XK 26 column containing 23 ml of Red 120 Sepharose 4B-CL was used; :/1 /2 mg of protein were applied to each milliliter of gel taking care that the conductivity of the sample was below the conductivity of the 50 mM potassium phosphate equilibration buffer (pH 7.0); elution was carried out with a three-step gradient at 0, 0.2 and 1 M NaCl in 50 mM potassium phosphate buffer, pH 7.2; gel filtration with Sephadex G 25 was used to remove salt prior to the re-chromatography on the dye ligand column and concentration of protein solutions was carried out by ultrafiltration with an Amicon stirred cell or Filtron centrifugal microconcentrator tubes and a molecular mass cut-off of 10 kDa. b
B. Nidetzky et al. / Chemico-Biological Interactions 143 /144 (2003) 533 /542
Fig. 1. Purification of GmXDH documented by SDS PAGE. Visualization of protein bands was by staining with Coomassie brilliant blue. Lanes (left to right), 1: molecular mass standards (in kDa); 2: cell extract; Red 120 chromatography; Red 120 rechromatography.
meters of GmXDH for NAD -dependent reactions with D-sorbitol and xylitol at pH 10.0 and NADH-dependent reactions with D-fructose and D-xylulose at pH 7.0 were identical to corresponding kinetic parameters measured with the natural enzyme (not shown). 3.2. Structural and functional properties of GmXDH The molecular mass of GmXDH subunit was determined by SDS-PAGE (Fig. 1) and is 38 (9/1) kDa. This is the expected size for the full-length protomer of xylitol dehydrogenase. In sizing chromatography using a Superose 12 HR 10/30 gelfiltration column, GmXDH eluted as a single protein peak with an apparent molecular mass of 160 kDa. Hence, similar to natural xylitol dehydrogenase, the recombinant enzyme is a functional tetramer and appears not to exist as a mixture of different oligomeric forms in solution. The pI of GmXDH is :/6.0, in good agreement with the pI value of the natural enzyme [3]. GmXDH contained 0.99/0.2 mole of Zn2 per each mole of 37.4-kDa protomer, indicating that zinc stoichiometry of the natural enzyme had been preserved completely during recombinant protein production. In contrast to the natural enzyme that was
537
found to contain 6 Mg2 per protein subunit, GmXDH preparations did not show a significant content of Mg2. We cannot offer an explanation for this difference on the basis of data available at the present time. The heterologous environment in E. coli as well as differences in protocols used for purification of natural and recombinant xylitol dehydrogenase could be relevant factors. The stereospecificity of hydrogen transfer to and from NADH was determined by 1H-NMR, essentially as described recently for Candida tenuis xylose reductase [12]. When 4-R [2H]-NADH was employed as co-substrate for reduction of Dfructose by GmXDH, the deuterium was depleted such that after complete oxidation of labeled NADH authentic NAD was obtained (not shown). By contrast, when 4-S [2H]-NADH was used, the hydrogen was depleted at the C4 position of the nicotinamide ring. The product revealed clearly by NMR analysis was [4-2H]-NAD . Therefore, GmXDH is specific for transferring the 4-pro-R hydrogen of NADH. The observed (‘A-side’) stereospecificity of hydrogen transfer is typical of members of the MDR superfamily [16]. 3.3. Kinetics Initial rates of NAD -dependent oxidation of xylitol and D-sorbitol (forward reaction) and NADH-dependent reduction of D-xylulose and D-fructose (reverse reaction) were recorded in the absence of products at pH 7.5 and 25 8C. In each direction of the reaction, they were acquired under conditions in which the substrate was varied at several constant levels of the coenzyme (Fig. 2a, b). Graphical analysis of the data for polyol oxidations in double reciprocal plots of the form 1/v versus 1/[polyol] gave intersecting patterns with crossover points that were to the left of the vertical axis and above the horizontal axis (Fig. 2a). The same analysis for initial rates of Dfructose reduction yielded a crossover point as far below the horizontal axis as it was above the same axis in the direction of D-sorbitol oxidation (not shown). Note that this is a requirement of the Theorell /Chance mechanism although it does not, of course, prove the mechanism [17]. The result is an indication, however, that the mechanism is not
538
B. Nidetzky et al. / Chemico-Biological Interactions 143 /144 (2003) 533 /542
Fig. 2. Double reciprocal plots of initial rates of NAD -dependent oxidation of xylitol (a) and NADH-dependent reduction of Dxylulose (b) by GmXDH measured at varied substrate concentrations and at several constant levels of co-substrate, as indicated in the plots. In the left-hand side panel (a), NAD concentrations were 0.1 (full squares), 0.2 (open triangles), 0.5 (full triangles), 1.0 (open circles) and 2.0 mM (full circles). In the right-hand side panel (b), NADH concentrations were 20 (open triangles), 50 (full triangles), 100 (open circles) and 200 mM (full circles). Data are for reactions in 50 mM potassium phosphate buffer, pH 7.5. The symbols show the mean values of two independent determinations and lines represent the fit of the shown data to Eq. (3).
rapid equilibrium random. Results for D-xylulose reduction gave initial rate patterns that were intersecting (Fig. 2b). However, the crossover point was not as clearly defined as in the opposite direction of reaction. According to the fit of the data to Eq. (3), it lay above the horizontal axis, as shown in Fig. 2(b). Cleland has discussed that in a sequential bi-reactant mechanism, where both substrates dissociate more slowly than they react to yield products, initial rate patterns may occur that are not clearly intersecting and may occasionally look nearly parallel [17]. Isotope effect data (not shown) suggest that both D-xylulose and NADH may be sticky indeed. The experimental data in Fig. 2 were fitted to Eq. (3) and with the assumption of ordered reactant binding, [A ] and [B ] are coenzyme and substrate, respectively, KiA is the apparent dissociation constant of enzyme / coenzyme complex and KA and KB are Michaelis constants.
v kcat [E][A][B]= (KiA KB KB [A]KA [B][A][B])
(3)
Estimates for the kinetic parameters are summarized in Table 2. The internal consistency of each set of kinetic parameters was checked with the Haldane relationship for a sequential bi-bi kinetic mechanism. The equilibrium constant (Keq) calculated by using kinetic parameters for reactions with xylitol and D-xylulose in Tris buffer was 5.1 /10 11 M. This value for Keq is in good agreement with the experimental equilibrium constant of 7 /10 11 M for the reaction, xylitol/ NAD l/D-xylulose/NADH/H [18]. Likewise, the calculated equilibrium constant of 2.9 /10 9 M for the enzymic reactions with Dsorbitol and D-fructose agrees with the value of 3.7 /10 9 M determined experimentally [19]. Keq for reactions with xylitol and D-xylulose in phos-
B. Nidetzky et al. / Chemico-Biological Interactions 143 /144 (2003) 533 /542
539
Table 2 Kinetic parameters of GmXDH at 25 8C and pH 7.5 Parameter
50 mM Tris /HCl buffer a D-Sorbitol
kcat (s 1) KiNAD (mM) KNAD (mM) Kpolyol (mM)
kcat (s 1) KiNADH (mM) KNADH (mM) Kketose (mM)
1149/2 7409/110 2279/18 559/4
50 mM Potassium phosphate buffer Xylitol
Xylitol
1439/4 4109/100 3209/32 399/4
1709/8 7809/194 1409/20 129/2
a D-Fructose
D-Xylulose
D-Xylulose
9209/120 239/9 (249/5)b 1609/30 13009/200
15009/200 279/6 869/8 109/1
18009/350 (1909/110)c 739/24 89/4
5.1 /10 11
4.8 /10 10
Haldane Keq (M)
2.9 /10 9
a Analysis of the crossover point for intersecting initial rate pattern can be carried out by using the relationship (1/v )X over / (kcat[E ]) 1 (1 /KNAD(H)/KiNAD(H)) [17] where (1/v )X over is the crossover point. b Determined from linear competitive inhibition of NADH against NAD . c This value for KiNADH should not be interpreted as an apparent dissociation constant of enzyme /NADH complex.
phate buffer was 4.8 /10 10 M and about seven times the value expected. Interpretation of the KiNAD(H) values in Table 2 as apparent dissociation constants for NADH and NAD requires that the kinetic mechanism be ordered with co-substrate binding first and leaving last. A value of 0.8 mM for KiNAD has recently been calculated from initial-rate data for NAD dependent oxidation of D-sorbitol in phosphate buffer by GmXDH [3]. It is in good agreement with the KiNAD values found here in both phosphate- and Tris-buffered solutions (Table 2), suggesting that KiNAD could reflect binding of NAD to the free enzyme. An estimate of 10 mM for KiNADH has been obtained from steady-state kinetic analysis of D-fructose reduction in phosphate buffer by GmXDH [3]. Equilibrium binding studies, in which NADH was used as inhibitor against co-substrate in the NAD -dependent oxidation of D-sorbitol, yielded a closely similar value of 12 mM for KiNADH [3]. Likewise, three independent estimates for KiNADH have been obtained herein using Tris-buffered solution. Though about 2.5 times higher than the corresponding KiNADH values measured in phosphate buffer, the values for KiNADH in Table 2 were
nearly identical (:/25 mM), as expected if NADH bound to the free enzyme under the conditions used. It is interesting, therefore, that steady-state kinetic analysis of D-xylulose reduction in phosphate buffer yielded a value of 190 mM for KiNADH, which is almost 20 times that determined in the previous study [3]. Although the value is illdefined statistically, it is very probably /25 mM. This result shows that KiNADH does not represent an apparent dissociation constant of the enzyme / NADH complex and might imply a random mechanism of reaction of GmXDH with D-xylulose and NADH. If binding of D-xylulose and NADH to GmXDH were completely random and rapiD-equilibrium conditions applied, the relationship KA/KiA /a could be used to calculate a value of 0.39 for a , which is the factor by which binding of one substrate changes the dissociation constant for the other substrate. The fact that a B/1 implies that the affinity for D-xylulose increases when NADH is bound and likewise the affinity for NADH increases when D-xylulose is bound. Synergism between co-substrate and substrate binding has been also observed with liver alcohol dehydrogenase [20,21].
540
B. Nidetzky et al. / Chemico-Biological Interactions 143 /144 (2003) 533 /542
Comparisons of kcat values for reductions of Dxylulose and D-fructose are interesting (Table 2 and Ref. [3]). Clearly, kcat for reduction of Dxylulose is :/1.6- to 2.3-fold higher than kcat for reduction of D-fructose. It has been shown recently that kcat for reduction of D-fructose in phosphate buffer, pH 7.5, is (completely) limited by dissociation of the enzyme /NAD complex. Release of NAD takes place with an estimated rate constant of 779 s 1 [3]. Now, if reaction of GmXDH in the presence of saturating concentrations of Dxylulose went through a compulsorily ordered mechanism involving enzyme /NAD complex, turnover number could not have a value larger than :/800 s 1, at variance with the experimental observations. The results imply that dissociation of NAD cannot be fully rate-limiting for turnover during D-xylulose reduction. A plausible scenario would be that an abortive enzyme / NAD /D-xylulose complex is formed at high Dxylulose levels which releases NAD between 1.6and 2.3-fold more rapidly than the enzyme / NAD complex. In that case, the binary complex with D-xylulose could be important kinetically and contribute to the distribution of the enzyme in the steady state. Protection afforded by 20 mM Dxylulose against inactivation of GmXDH by metal chelators such as EDTA suggests that D-xylulose is capable of binding to the free enzyme (not shown). In the literature on MDRs, there is precedence for kinetic mechanisms involving abortive complexes Table 3 Kinetic isotope effects on kinetic parameters for NADHdependent reduction of D-fructose and NAD -dependent oxidation of D-sorbitol of GmXDH, determined in 50 mM potassium phosphate buffer, pH 7.5, and at 25 8C Isotope effect
Reduction
D
n.a.a 1.829/0.25 1.209/0.16
kcat kcat/Kfructose D kcat/KNADH D
Oxidation D
kcat kcat/Ksorbitol D kcat/KNAD D
a
1.109/0.10 1.939/0.20 1.209/0.21
n.a., Not applicable because saturation in substrate could not be achieved.
and kinetically significant enzyme /substrate complexes [22 /25]. For example, an enzyme /NADH / alcohol complex is probably physiologically significant in reaction catalyzed by human liver alcohol dehydrogenase g2 [24]. In an alternative interpretation of our data, however, release of NAD could occur from the enzyme /xylitol / NAD complex, implying a random component to the mechanism. Clearly, more data will be required to establish a complete kinetic mechanism of GmXDH for the physiological reaction, xylitol/NAD l/D-xylulose/NADH/H . 3.4. Kinetic isotope effects Isotope effects on kinetic parameters for NADH-dependent reduction of D-fructose were measured by comparing initial rates recorded in 50 mM phosphate buffer, pH 7.5, and at 25 8C in the presence of unlabeled co-substrate and 4-R [2H]NADH. Data were collected under conditions in which either co-substrate was constant at 220 mM (:/3 /KNADH) or substrate was constant at a level of 1.5 M (:/KB for D-fructose). Data were fitted to Eq. (2) and results are shown in Table 3. Kinetic isotope effects on kinetic parameters for oxidation of D-sorbitol were determined under the above described buffer and temperature conditions from a comparison of initial rates recorded in the presence of unlabeled substrate and 2-[2H]-labeled polyol. Data measured at constant co-substrate concentration (3 mM; /5 /KNAD) or constant substrate concentration (300 mM; /5 /Kpolyol) were fitted to Eq. (2) and isotope effects are shown in Table 3. In the direction of D-fructose reduction, Dkcat/ Kfructose /Dkcat/KNADH and therefore, the release of D-fructose from the central complex must take place at a rate faster than the rate of release of NADH. The value for Dkcat/KNADH contains uncertainty because saturation in D-fructose could not be achieved in the experiment (Kfructose /1 M). (An isotope effect on kcat is, therefore, not available.) If the mechanism of GmXDH were strictly ordered and NADH was the substrate binding first, no isotope effect would be expected on kcat/KNADH in the presence of saturating concentrations of the second substrate. In that
B. Nidetzky et al. / Chemico-Biological Interactions 143 /144 (2003) 533 /542
case, the commitment of enzyme /NADH to undergo reaction as opposed to dissociate would become infinite and thus the isotope effect on the hydride transfer step would be completely masked [26]. The level of the second substrate will affect the magnitude of the commitment to catalysis and the concentration of D-fructose required to halfsuppress the isotope effect on kcat/KNADH would be given by the expression KiNADHKfructose/ KNADH. Taking data from Lunzer et al. [3], we calculate this concentration to be 0.32 M, whereas the actual concentration used in the experiment was 1.5 M. Considering that the value of Dkcat/ KNADH is actually close to and perhaps statistically not different from 1, the results agree with a mechanism of NADH-dependent reduction of Dfructose, in which reactant binding is (preferentially) ordered. The isotope effect on kcat for D-sorbitol oxidation has a value very close to 1.00, suggesting that product dissociation is completely rate-limiting for the overall reaction at pH 7.5. As shown in Table 3, Dkcat/Ksorbitol /Dkcat/KNAD and Dkcat/KNAD has a small value of 1.2 that perhaps is not statistically different from 1.0. This result indicates that although NAD might be released from the ternary complex at a finite, however small rate, the rate of dissociation of NAD is clearly very much slower than the corresponding dissociation rate of D-sorbitol. In conclusion, the isotope effect data for D-sorbitol oxidation and D-fructose reduction by GmXDH at pH 7.5 suggest an ordered kinetic mechanism of NAD -dependent D-sorbitol oxidation in which NAD binds before D-sorbitol and NADH is released only after Dfructose has dissociated. The rate of last product release is rate-limiting for kcat in this direction. The results, therefore, provide evidence supporting the Theorell /Chance mechanism proposed recently [3].
Acknowledgements Financial support from the Austrian Science Funds is gratefully acknowledged (FWF projects P-12569-MOB and P-15208-MOB to B.N.)
541
References [1] B. Hahn-Ha¨gerdal, H. Jeppson, K. Skoog, B.A. Prior, Biochemistry and physiology of xylose fermentation by yeasts, Enzyme Microb. Technol. 16 (1994) 933 /943. [2] B. Hahn-Ha¨gerdal, C.F. Wahlbom, M. Gardonyi, W.H. van Zyl, R.R. Cordero Otero, L.J. Jonsson, Metabolic engineering of Saccharomyces cerevisiae for xylose utilization, Adv. Biochem. Eng. Biotechnol. 73 (2001) 53 /84. [3] R. Lunzer, Y. Mamnun, D. Haltrich, K.D. Kulbe, B. Nidetzky, Structural and functional properties of a yeast xylitol dehydrogenase, a Zn2 -containing metalloenzyme similar to medium-chain sorbitol dehydrogenases, Biochem. J. 336 (1998) 91 /99. [4] A. Habenicht, H. Motejadded, M. Kiess, A. Wegerer, R. Mattes, Xylose utilisation: cloning and characterisation of the xylitol dehydrogenase from Galactocandida mastotermitis , Biol. Chem. 380 (1999) 1405 /1411. [5] B. Persson, J.S. Zigler, H. Jo¨rnvall, A super-family of medium-chain dehydrogenases/reductases (MDR). Sublines including z-crystallin, alcohol and polyol dehydrogenases, quinone oxidoreductases, enoyl reductases, VAT-1 and other proteins, Eur. J. Biochem. 226 (1994) 15 /22. [6] C. Karlsson, J.-O. Ho¨o¨g, Zinc coordination in mammalian sorbitol dehydrogenase, Eur. J. Biochem. 216 (1993) 103 / 107. [7] J. Jeffery, H. Jo¨rnvall, Sorbitol dehydrogenase, Adv. Enzymol. 61 (1988) 47 /106. [8] M.J. Banfield, M.E. Salvucci, E.N. Baker, C.A. Smith, Crystal structure of the NADP(H)-dependent ketose ˚ resolution, J. reductase from Bemisia argentifolii at 2.3 A Mol. Biol. 306 (2001) 239 /250. [9] K. Johansson, M. El-Ahmad, C. Kaiser, H. Jo¨rnvall, H. Eklund, J.-O. Ho¨o¨g, S. Ramaswamy, Crystal structure of sorbitol dehydrogenase, Chem.-Biol. Interact. 130 /132 (2001) 351 /358. [10] H. Eklund, E. Horjales, H. Jo¨rnvall, C.-I. Bra¨nde´n, J. Jeffery, Molecular aspects of functional differences between alcohol and sorbitol dehydrogenases, Biochemistry 24 (1985) 8005 /8012. [11] P. Mayr, K. Bru¨ggler, K.D. Kulbe, B. Nidetzky, Xylose metabolism in Candida intermedia : efficient isolation and characterisation of two aldose reductases with different coenzyme specificities, J. Chromatogr. B 737 (2000) 195 / 202. [12] B. Nidetzky, P. Mayr, W. Neuhauser, M. Puchberger, Structural and functional properties of aldose (xylose) reductase from the D-xylose-metabolizing yeast Candida tenuis , Chem.-Biol. Interact. 130 /132 (2001) 583 /595. [13] M. Klimacek, B. Nidetzky, Examining the relative timing of hydrogen abstraction steps during NAD -dependent oxidation of secondary alcohols catalyzed by long-chain mannitol dehydrogenase from Pseudomonas fluorescens using pH and isotope effects, Biochemistry 41 (2002) 10158 /10165.
542
B. Nidetzky et al. / Chemico-Biological Interactions 143 /144 (2003) 533 /542
[14] G. Mayer, K.D. Kulbe, B. Nidetzky, Utilization of xylitol dehydrogenase in a combined microbial/enzymatic process for production of xylitol from D-glucose, Appl. Biochem. Biotechnol. 98 /100 (2002) 577 /589. [15] D.B. Northrop, Deuterium and tritium kinetic isotope effects on initial rates, Methods Enzymol. 87 (1982) 607 / 625. [16] R. Meijers, R.J. Morris, H.W. Adolph, A. Merli, V.S. Lamzin, E.S. Cedergren-Zeppezauer, On the enzymatic activation of NADH, J. Biol. Chem. 276 (2001) 9316 / 9321. [17] W.W. Cleland, Determining the chemical mechanisms of enzyme-catalyzed reactions by kinetic studies, Adv. Enzymol. 45 (1977) 273 /387. [18] M. Rizzi, K. Harwart, N.-A. Bui-Thanh, H. Dellweg, A kinetic study of the NAD -xylitol dehydrogenase from the yeast Pichia stipitis , J. Ferment. Bioeng. 67 (1989) 25 / 30. [19] R.I. Lindstad, L.F. Hermansen, J. McKinley-McKee, The kinetic mechanism of sheep liver sorbitol dehydrogenase, Eur. J. Biochem. 210 (1992) 641 /647. [20] P. Andersson, J. Kvassman, B. Olde´n, G. Pettersson, Synergism between coenzyme and aldehyde binding to
[21]
[22]
[23]
[24]
[25]
[26]
liver alcohol dehydrogenase, Eur. J. Biochem. 133 (1983) 651 /655. P. Andersson, J. Kvassman, B. Olde´n, G. Pettersson, Synergism between coenzyme and alcohol binding to liver alcohol dehydrogenase, Eur. J. Biochem. 144 (1984) 317 / 324. P. Andersson, J. Kvassman, B. Olde´n, G. Pettersson, Catalytic significance of binary enzyme /aldehyde complexes in the liver alcohol dehydrogenase reaction, Eur. J. Biochem. 139 (1984) 519 /527. F.M. Dickinson, G.P. Monger, A study of the kinetics and mechanism of yeast alcohol dehydrogenase with a variety of substrates, Biochem. J. 131 (1973) 261 /270. H.A. Charlier, Jr, B.V. Plapp, Kinetic cooperativity of human liver alcohol dehydrogenase g2, J. Biol. Chem. 275 (2000) 11569 /11575. K. Dalziel, F.M. Dickinson, Substrate activation and inhibition in coenzyme /substrate reactions, Biochem. J. 100 (1966) 491 /500. P.F. Cook, Kinetic and regulatory mechanisms of enzymes from isotope effects, in: P.F. Cook (Ed.), Enzyme Mechanisms From Isotope Effects, CRC Press, Boca Raton, 1991, p. 203.